> ## Documentation Index
> Fetch the complete documentation index at: https://docs.workshop.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Using Credits Effectively

> How to manage conversations, project context, caching, and model choice without wasting credits.

The simplest way to avoid unnecessary credit use is to start a new conversation when
the goal changes or you return after a long break.

<Info>
  **When should you leave a long conversation behind?** When you begin a different task,
  or return after more than an hour of inactivity—even if you are continuing the same
  work. The usual cache window is about one hour, so reopening an older conversation can
  mean paying for the model to process its full history again. Start a new conversation,
  or switch to a lower-cost model, use `/compact`, then return to the capable model if
  you need to continue the same task.
</Info>

## What should I do?

### While actively working

| Situation                                                              | Best action                                                                                                                                                         |
| ---------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Continuing the same focused task while the cache is warm               | Keep the same conversation and model                                                                                                                                |
| Starting a different feature, bug, or topic                            | Start a new conversation; before leaving the current one, run `/context` if it contains lasting knowledge                                                           |
| Continuing the same task without a long break, but with a long history | Run `/compact` to shorten the history carried into later turns; if needed, say what the summary should preserve                                                     |
| Changing the main model                                                | If needed, first run `/context` to save lasting knowledge or `/compact` to continue the same task; then select the new model in the fresh or compacted conversation |
| The agent is repeating unsuccessful attempts                           | Interrupt it, narrow the goal, or return to planning                                                                                                                |

### Before taking a long break

Prepare the handoff while the cache is still warm.

| What you need later                             | Best action                                                              |
| ----------------------------------------------- | ------------------------------------------------------------------------ |
| Knowledge that future conversations should know | Run `/context`; if needed, say what project knowledge it should preserve |
| The ability to continue the same task           | Run `/compact` to create a shorter continuation                          |

You can view or edit the saved project context in `.workshop/context.md`.

If both apply, run `/context` first, then `/compact`.

### When you come back

| Situation                                                        | Best action                                                                                                                                                                                                                                               |
| ---------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| You saved the important knowledge with `/context` before leaving | Start a new conversation and state the next goal                                                                                                                                                                                                          |
| You shortened the history with `/compact` before leaving         | Continue from that shortened conversation                                                                                                                                                                                                                 |
| The old conversation is cold but contains unsaved knowledge      | **Start fresh:** switch to a lower-cost model, run `/context`, then use the capable model in a new conversation.<br /><br />**Continue the task:** switch to a lower-cost model, run `/compact`, then use the capable model in the shortened conversation |
| You do not need anything from the old conversation               | Start a new conversation instead of continuing it                                                                                                                                                                                                         |

## Three things that affect credit use

|                          | What it is                                                              | How to manage it                                                                                      |
| ------------------------ | ----------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| **Project context**      | Stable knowledge shared across conversations                            | Use `/context` to save important project knowledge in `.workshop/context.md` for future conversations |
| **Conversation context** | Messages, file reads, commands, and agent responses in one conversation | Keep it focused; use `/compact` to shorten the history carried into later turns, or start fresh       |
| **Prompt cache**         | A temporary copy of input the model has already processed               | Keep the same model while the cache is warm; assume it is cold after a long break                     |

## Why long conversations cost more

Each new turn includes the context needed to continue the conversation. As messages, file
reads, and tool results accumulate, there is more history to process on every turn.

A warm prompt cache makes repeated history cheaper, but it does not make the conversation
smaller. Output, new input, agent turns, and delegated work still consume credits.

<Info>
  **Why a fresh conversation saves credits.** It leaves the old chat history behind while
  keeping the project's files, sources, and saved project context. The agent starts with
  the durable information it needs, without carrying every earlier message and tool result.
</Info>

Use one conversation for one clear purpose. When the purpose changes, start a new one.

## Use `/context` to save knowledge for future conversations

`/context` saves important project knowledge from the current conversation in
`.workshop/context.md` so future conversations can use it.

```text theme={null}
/context focus on architecture decisions, the current migration status,
and the commands used to verify the project
```

Run it after a milestone or lasting decision. Keep the saved context concise, and never
include secrets. You can view or edit it in `.workshop/context.md`. See
[Context and Memory](/core-concepts/context-and-memory).

## Use `/compact` to shorten history for the same task

`/compact` replaces a long working history with a shorter summary. Use it when you want to
continue the same task but no longer need every intermediate step.

```text theme={null}
/compact focus on the authentication bug, attempted fixes, and remaining verification
```

Compaction requires a summarization turn and usually starts a new cache. The saving comes
from carrying the shorter summary on later turns.

Workshop also compacts automatically near the context limit. See
[Working with the Agent](/core-concepts/working-with-the-agent#auto-compaction).

## Why caching and model switching matter

A prompt cache lets a model reuse conversation history at a discounted rate. The usual
TTL is **one hour**. Think of this as an idle timer: using the cache can refresh it, but
the exact behavior varies by model and provider.

| What happens                                         | Result                                         |
| ---------------------------------------------------- | ---------------------------------------------- |
| Continue with the same model while the cache is warm | Matching history can use the lower cache rate  |
| Return after the cache expires                       | The model must process the history again       |
| Switch the main model or provider                    | The new model cannot use the old model's cache |

<Info>
  **Why a cold premium turn can be surprisingly expensive.** Think of two prices
  multiplying. A premium model can cost around **10× more** than a lower-cost model. If
  the cache is cold, rereading the same long conversation can cost around **10× more**
  than reading it from a warm cache. Together, the conversation-history part of the first
  cold turn on a premium model can cost around **100× more** than a warm turn on a
  lower-cost model.

  The exact difference varies, and the 100× comparison does not apply to the entire turn.
  The practical rule is simple: do not make a premium model process a long, cold history.
  Use a lower-cost model for `/context` or `/compact`, then use the capable model in the
  fresh or shortened conversation.
</Info>

## Choose the model for the task

The cheapest turn is not always the cheapest completed task. Repeated corrections can
cost more than using a stronger model once.

| Work                                                  | Model strategy                      |
| ----------------------------------------------------- | ----------------------------------- |
| Routine edits, searches, and straightforward tasks    | Use Auto or a lower-cost model      |
| Everyday implementation                               | Use a balanced model                |
| Difficult debugging, migrations, or architecture work | Choose a premium model deliberately |

[Auto](/core-concepts/settings#auto) is the simplest default when you do not want to pick
a model yourself.

## Related reading

* [Context and Memory](/core-concepts/context-and-memory)
* [When to Start Fresh vs. Persist](/troubleshooting/when-to-start-fresh)
* [Working with the Agent](/core-concepts/working-with-the-agent)
* [Plans and credits](/pricing/plans)